Battery early degradation early warning method and device, vehicle and storage medium

Abstract

The application relates to a battery early degradation early warning method and device, a vehicle and a storage medium, wherein the method comprises the following steps: acquiring voltage values of all single batteries in a battery pack at the same time to form a voltage sample set; calculating a plurality of statistical characteristic quantities in the voltage sample set; constructing a degradation characteristic sequence in time sequence according to the plurality of statistical characteristic quantities obtained at different detection times of the same battery pack; extracting a change slope of each statistical characteristic quantity in the degradation characteristic sequence by using a sliding window linear regression, and calculating a comprehensive degradation score based on the change slope; comparing the comprehensive degradation score with a preset grading threshold value, and comparing each change slope with a single-dimensional trend threshold value to determine and output an early warning level. The application can timely capture early signs of battery degradation, and improve the timeliness and accuracy of early warning.

Description

Technical Field

[0001] This application relates to the field of battery health status monitoring technology, and in particular to a method, device, vehicle, and storage medium for early warning of early battery degradation. Background Technology

[0002] With the rapid popularization of new energy vehicles, the health status assessment of power batteries has become a core aspect of ensuring safe vehicle operation. Currently, the widely used assessment methods in the industry primarily rely on the battery management system (BMS) to collect voltage data from all individual cells within the battery pack. From this data, the highest and lowest voltage values ​​are selected, and extreme values ​​such as the maximum voltage difference are calculated. These values ​​are then compared with preset, fixed empirical thresholds to determine abnormal conditions. For example, when the maximum voltage difference exceeds 50 millivolts, the system marks it as "caution," and when it exceeds 100 millivolts, it is upgraded to an "abnormal" level.

[0003] However, existing methods reveal a key flaw in practical applications: a severe lag in the detection process. Essentially, they are a reactive mechanism, triggering an alarm only when cell consistency within the battery pack significantly deteriorates and the voltage difference reaches a high threshold. During battery degradation, there is a long nascent degradation period between the initial healthy state and significant voltage difference exceeding the standard. During this stage, minute voltage differences between cells accumulate and gradually increase, but because the magnitude of these changes has not yet reached a fixed extreme threshold, existing technologies are completely unable to capture such early abnormal trends. This means that by the time the alarm is finally activated, the battery has often already suffered irreversible performance damage, and the optimal time for maintenance and intervention has long been lost. Furthermore, existing methods overemphasize extreme values ​​such as the highest and lowest voltages, ignoring the voltage distribution information of the majority of intermediate cells within the battery pack. They fail to effectively reflect overall characteristics such as voltage dispersion, distribution symmetry, and concentration trends, making the system extremely insensitive to subtle, gradual changes in battery consistency, further amplifying the detection delay problem. Therefore, there is an urgent need for a technology that can accurately identify and warn of abnormal trends in the early stages of battery degradation, i.e., when the voltage difference is far below the traditional alarm threshold. Summary of the Invention

[0004] The purpose of this application is to provide a method, device, vehicle, and storage medium for early warning of battery degradation, which can promptly capture early signs of battery degradation and improve the timeliness and accuracy of the warning.

[0005] To address the aforementioned technical problems, embodiments of this application provide a method for early warning of battery degradation, comprising: Obtain the voltage values ​​of all individual cells in the battery pack at the same time to form a voltage sample set; Calculate multiple statistical features in the voltage sample set; For multiple statistical features obtained at different detection times for the same battery pack, a degradation feature sequence is constructed in chronological order. The slope of change of each statistical feature in the degradation feature sequence is extracted by sliding window linear regression, and a comprehensive degradation score is calculated based on the slope of change. The comprehensive degradation score is compared with a preset grading threshold, and each of the change slopes is compared with a single-dimensional trend threshold to determine and output the warning level.

[0006] To address the aforementioned technical problems, embodiments of this application provide a battery early degradation warning device, comprising: The voltage sample set acquisition module is used to acquire the voltage values ​​of all individual cells in the battery pack at the same time to form a voltage sample set. The statistical feature calculation module is used to calculate multiple statistical features in the voltage sample set; The degradation feature sequence construction module is used to construct a degradation feature sequence in chronological order for multiple statistical feature quantities obtained at different detection times for the same battery pack. The comprehensive degradation score calculation module is used to extract the slope of change for each statistical feature quantity in the degradation feature sequence using sliding window linear regression, and to calculate the comprehensive degradation score based on the slope of change. The warning level output module is used to compare the comprehensive degradation score with the preset grading threshold, and to compare each of the change slopes with the single-dimensional trend threshold, so as to determine and output the warning level.

[0007] To solve the above-mentioned technical problems, one technical solution adopted by the present invention is to provide a vehicle, including one or more processors; and a memory for storing one or more programs, such that the one or more processors implement the battery early degradation warning method described in any one of the above-mentioned methods.

[0008] To solve the above-mentioned technical problems, one technical solution adopted by the present invention is: a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the battery early degradation warning method described in any one of the above-mentioned methods.

[0009] This invention provides a method, device, vehicle, and storage medium for early warning of battery degradation. By acquiring the voltage of all individual cells within the battery pack to construct a sample set, multi-dimensional statistical features are calculated and a time-series degradation feature sequence is constructed. Combined with sliding window linear regression to extract the dynamic change slope and comprehensive degradation score, accurate identification and graded early warning of battery degradation are achieved. This effectively overcomes the lag defect caused by existing technologies relying solely on extreme value indicators. It can promptly capture early signs of battery degradation, providing warnings even when the voltage difference is far below the traditional alarm threshold, significantly improving the timeliness and accuracy of warnings. Through multi-dimensional statistical features and dynamic trend analysis, the overall changing characteristics of battery voltage distribution are comprehensively reflected, effectively reducing the risk of false alarms and missed alarms. Attached Figure Description

[0010] To more clearly illustrate the solutions in this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0011] Figure 1 This is a flowchart of an implementation of the battery early degradation early warning method provided in this application embodiment; Figure 2 This is a flowchart illustrating the implementation of the first sub-process in the battery early degradation warning method provided in this application embodiment; Figure 3 This is a schematic diagram illustrating the calculation of statistical characteristics of cell voltage distribution provided in an embodiment of this application; Figure 4 This is a flowchart illustrating the implementation of the second sub-process in the battery early degradation warning method provided in this application embodiment; Figure 5 This is a flowchart illustrating the implementation of the third sub-process in the battery early degradation warning method provided in this application embodiment; Figure 6 This is a flowchart illustrating the implementation of the fourth sub-process in the battery early degradation warning method provided in this application embodiment; Figure 7 This is a schematic diagram of the battery early degradation early warning device provided in the embodiments of this application; Figure 8 This is a schematic diagram of the vehicle provided in the embodiments of this application. Detailed Implementation

[0012] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of this application; the terms "comprising" and "having," and any variations thereof, in the specification, claims, and foregoing drawings of this application, are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the specification, claims, or foregoing drawings of this application are used to distinguish different objects, not to describe a particular order.

[0013] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in different places throughout this specification does not necessarily refer to the same embodiment, nor is it intended to be an independent or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0014] Existing methods for assessing the health status of power batteries primarily involve collecting extreme values ​​such as the highest, lowest, and maximum voltage difference of individual cells within the battery pack and comparing them with fixed empirical thresholds. However, this method suffers from significant lag, triggering alarms only when battery consistency deteriorates severely and the voltage difference reaches a high threshold, thus missing the opportunity for early intervention. Furthermore, it ignores the voltage distribution information of the majority of intermediate cells, making it difficult to accurately capture subtle, gradual changes in the overall consistency of the battery pack.

[0015] To address this issue, this application proposes a method for early warning of battery degradation. This method involves acquiring the voltage values ​​of all individual cells within a battery pack at the same time, forming a voltage sample set; calculating multiple statistical features within the voltage sample set; constructing a degradation feature sequence in chronological order for the multiple statistical features obtained at different detection times for the same battery pack; extracting the slope of change for each statistical feature in the degradation feature sequence using sliding window linear regression; and calculating a comprehensive degradation score based on the slope of change; comparing the comprehensive degradation score with a preset grading threshold, and comparing each slope of change with a single-dimensional trend threshold, to determine and output the warning level. This method aims to overcome the detection lag problem in existing technologies and achieve early and sensitive warning of battery degradation.

[0016] For ease of understanding, the following explains some key terms in this embodiment: A voltage sample set refers to the collection of voltage values ​​collected from all individual cells within a battery pack at a specific moment. This set contains real-time voltage information for all cells in the battery pack and serves as the foundational data for subsequent analysis.

[0017] Statistical characteristics are quantitative indicators extracted from voltage sample sets that reflect the overall voltage distribution characteristics, dispersion, or central tendency of the battery pack. These indicators are used to transform raw, large amounts of voltage data into more representative values ​​that can be used for trend analysis.

[0018] A degradation feature sequence refers to a sequence of multiple statistical features obtained at different detection times for the same battery pack, arranged in chronological order. This sequence reflects the state evolution trajectory of the battery pack over time.

[0019] Sliding window linear regression is a time series analysis method that extracts local trend information by moving a fixed-size window across the data series and performing linear regression analysis within each window.

[0020] The slope of change refers to the rate at which a statistical characteristic quantity changes over time after analyzing the statistical characteristic quantity in a degenerate characteristic sequence using a sliding window linear regression method. It reflects the upward or downward trend of the statistical characteristic quantity within a local time period.

[0021] The comprehensive degradation score is a holistic index calculated using a fusion algorithm based on the slopes of changes in multiple statistical features. This score aims to comprehensively quantify the overall degradation level of the battery pack.

[0022] Preset grading thresholds refer to pre-defined numerical limits used to grade the overall degradation score. By comparing the overall degradation score with these thresholds, the degradation stage of the battery pack can be determined.

[0023] Single-dimensional trend thresholds refer to the judgment limits set individually for the slope of change of each statistical characteristic. These thresholds are used to assess whether the trend of change of a single statistical characteristic exceeds the normal range.

[0024] The warning level refers to the final health status level of the battery pack, determined by comparing the comprehensive degradation score with preset grading thresholds, and comparing the slopes of change with single-dimensional trend thresholds. The warning level output aims to provide users or the system with a clear indication of the battery's health status.

[0025] This application overcomes the limitations of traditional methods that rely solely on extreme value indicators by comprehensively acquiring the voltage values ​​of all individual cells within the battery pack and calculating multiple statistical characteristics. By constructing a degradation feature sequence and employing sliding window linear regression to extract the slope of change, this method can keenly capture the accumulation of minute differences between cells and the subtle, gradual trend of overall consistency changes in the early stages of battery degradation. Therefore, it can provide a sensitive early warning of early battery degradation before irreversible damage occurs, effectively solving the problem of detection lag in existing technologies and providing a basis for timely battery maintenance and intervention.

[0026] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0027] Please see Figure 1 , Figure 1 This paper illustrates a specific implementation of a method for early warning of battery degradation.

[0028] It should be noted that if substantially the same result is obtained, the method of this invention is not based on... Figure 1 Limited to the order of the processes shown, this method includes the following steps: S1: Obtain the voltage values ​​of all individual cells in the battery pack at the same time to form a voltage sample set.

[0029] Specifically, the voltage values ​​of all individual cells within the battery pack at the same time are acquired to form a voltage sample set. This can be achieved through the voltage acquisition module of the battery management system (BMS). For example, the BMS can periodically sample the voltage of all series-connected cells within the battery pack and aggregate these voltage data to form a set containing the voltages of all individual cells. Alternatively, external diagnostic equipment can be connected to the battery pack to manually or automatically trigger a full-cell voltage acquisition, thereby obtaining the voltage sample set.

[0030] In one specific embodiment, the voltage values ​​V1, V2, ..., Vn of all individual cells in the battery pack are periodically acquired (e.g., every 10 minutes or at the end of each charge) via a vehicle-mounted acquisition module. The acquired raw voltage data is then uploaded to a processing unit via an in-vehicle communication network (e.g., CAN bus) or a wireless network. This processing unit can be an embedded controller in the vehicle or a server in the cloud. In another specific embodiment, the voltage sample set described in this application is acquired under conditions where the batteries are in the same or similar SOC range (e.g., SOC ∈ [40%, 60%]) and the temperature fluctuation does not exceed ±5℃, in order to eliminate the interference of charge / discharge state and temperature on the voltage distribution pattern. If the above conditions cannot be met, the voltage of each cell can be normalized based on the SOC-voltage mapping curve before calculating statistical characteristics.

[0031] S2: Calculate multiple statistical features in the voltage sample set.

[0032] Specifically, after obtaining the voltage sample set, several statistical characteristics need to be calculated from this sample set. These statistical characteristics are designed to reflect the distribution characteristics of cell voltage within the battery pack from different dimensions.

[0033] Please see Figure 2 and Figure 3 , Figure 2 One specific implementation of step S2 is shown. Figure 3 This is a schematic diagram illustrating the calculation of statistical characteristics of cell voltage distribution provided in an embodiment of this application, detailed below: S21: Sum the voltage values ​​of all cells in the voltage sample set and divide by the total number of cells to calculate the sample mean. S22: Based on the sample mean, calculate the variance reflecting the degree of voltage dispersion, the skewness reflecting the symmetry of the voltage distribution, and the kurtosis reflecting the degree of concentration of the voltage distribution.

[0034] Specifically, after obtaining the voltage values ​​of all individual cells within the battery pack at the same time to form a voltage sample set, statistical analysis of this voltage sample set is required to extract distribution characteristics that characterize the overall consistency of the battery pack. First, the sample mean is calculated. The sample mean is a statistical indicator that measures the central tendency of a set of data, representing the sum of all data points divided by the total number of data points. Its function is to provide a benchmark reference for the battery pack voltage, thereby measuring the overall voltage level and providing a unified scale for subsequent analysis of cell voltage fluctuations. For example, the sample mean can be obtained by iterating through all cell voltage values ​​in the voltage sample set, summing them, and then dividing by the total number of cells; alternatively, existing statistical computing libraries or software tools can be used to directly call their provided mean calculation functions to process the voltage sample set.

[0035] Building upon this, variance, reflecting the degree of voltage dispersion, is further calculated. Variance is an indicator of data dispersion, representing the average of the squares of the differences between each data point and the mean. By calculating variance, the dispersion of cell voltage can be effectively quantified, thus keenly capturing early signals of deteriorating consistency between cells. Simultaneously, skewness, reflecting the symmetry of voltage distribution, is calculated. Skewness is an indicator of data distribution asymmetry, describing the direction and degree of skewness of the tail of the data distribution relative to the mean. Positive skewness indicates a longer right tail in the data distribution, while negative skewness indicates a longer left tail. By introducing skewness, it is possible to reveal whether there are abnormal deviations from the mean in cell voltage from the perspective of voltage distribution symmetry, which is of great significance for identifying the early degradation of performance in some cells.

[0036] In addition, kurtosis, which reflects the concentration of voltage distribution, is calculated. Kurtosis is an indicator of the concentration or thickness of the tails in data distribution, describing the sharpness of the peaks and the thickness of the tails. High kurtosis indicates that the data is concentrated near the mean and has thick tails; low kurtosis indicates that the data distribution is relatively flat and has thin tails. By introducing kurtosis, the concentration of voltage distribution can be reflected, that is, it can be identified whether extreme voltage values ​​impact the overall distribution pattern, thus allowing potential performance problems to be detected in the early stages of degradation through changes in the distribution pattern.

[0037] In one specific embodiment, the statistical features are calculated as follows:

[0038] in, For variance, This represents the total number of battery cells. The sample mean. Let i be the voltage of the i-th cell. For skewness, The kurtosis is used in this embodiment as the original kurtosis, with a theoretical value of 3 for a normal distribution.

[0039] The multi-dimensional statistical feature system of this application enables the early warning system to extract more representative degradation features from the evolution of the overall distribution. Thus, in the early stage of battery degradation, that is, before the voltage difference reaches the traditional alarm threshold, it can sensitively capture the nonlinear changes and abnormal trends of cell voltage distribution, significantly improving the ability to capture early battery degradation trends and the sensitivity of the early warning system, and effectively solving the problem of detection lag in the prior art.

[0040] S3: For the same battery pack, multiple statistical features obtained at different detection times are used to construct a degradation feature sequence in chronological order.

[0041] Specifically, to capture the dynamic degradation process of the battery, the statistical features obtained from each detection need to be associated with the corresponding detection timestamp. These timestamped statistical features are then stored and arranged in ascending order of timestamps, forming a sequence reflecting the historical evolution of the battery pack. As a preferred implementation, the system can periodically and automatically record these statistical features and append them to a historical dataset, which is naturally arranged chronologically to form a degradation feature sequence.

[0042] Further, step S3 includes: marking the detection time for each group of statistical features; and arranging the historical detection records of the same battery pack in ascending order according to the detection time to construct the degradation feature sequence.

[0043] Specifically, the detection time is marked for each set of statistical features, aiming to assign a clear timestamp to each collected battery pack statistical feature data, thereby establishing a correlation between data and time. This is crucial for subsequent analysis of battery performance trends over time, ensuring the temporal integrity and traceability of the data. For example, the system clock can automatically record the current time when data acquisition is complete and store it as metadata along with the set of statistical features, such as using a Unix timestamp or a date and time string in ISO 8601 format. Alternatively, during data transmission or processing, the receiving end or processing module can append a detection time to each set of statistical features based on the arrival time of the data packet or internally recorded time information, for example, by including time information in the data packet header or automatically recording it through the insertion time field of a database.

[0044] Subsequently, the historical detection records of the same battery pack are arranged in ascending order according to the detection time. The purpose is to organize and sort the discrete statistical characteristic data of a specific battery pack acquired at different time points according to the chronological order of their detection times, forming a continuous, time-logical sequence. This provides correct data input for subsequent trend analysis (such as sliding window linear regression), ensuring that the analysis results accurately reflect the battery degradation process. For example, when storing data, a time series database (TSDB) or a regular database with a time index can be used. When querying or retrieving data, the ascending order can be directly specified by the detection time field. Alternatively, when the data is loaded into memory for processing, a sorting algorithm provided by the programming language (such as quicksort, merge sort, etc.) can be used, with the detection time as the sorting key, to arrange the record set containing statistical characteristics and detection time in ascending order.

[0045] In one specific embodiment, multiple sets of statistical feature vectors acquired from the same vehicle at different detection time points are arranged in chronological order to form a degradation feature sequence. Each set of feature vectors is marked with a timestamp, forming a multidimensional time series. ; Where maxDiff is the maximum pressure difference (as a benchmark).

[0046] The slope of change and the comprehensive degradation score calculated in the embodiments of this application will more reliably indicate the actual degradation trajectory of the battery, thereby enabling sensitive capture of early and subtle degradation patterns of the battery, significantly improving the timeliness and accuracy of early battery degradation warnings.

[0047] S4: For each statistical feature quantity in the degradation feature sequence, a sliding window linear regression is used to extract the slope of change, and a comprehensive degradation score is calculated based on the slope of change.

[0048] Specifically, after constructing the degradation feature sequence, a sliding window linear regression method needs to be applied to each statistical feature in the sequence to extract its slope. For example, a fixed-size sliding window can be set for each statistical feature. Within this window, linear regression analysis is performed on the data points in the window to obtain the slope of a fitted line, which is the slope of the statistical feature within the current window. The window slides forward along the time series, repeating this process to obtain a series of slopes. Based on these extracted slopes, a comprehensive degradation score can be calculated. For example, the various slopes can be weighted and summed to obtain a single comprehensive index that reflects the overall degradation trend.

[0049] It should be noted that when the number of data points in the degradation feature sequence is less than the amount of data required by the sliding window, the factory calibration parameters of the same model battery pack are used as the initial benchmark, or the normal level is directly output until the amount of data meets the window requirements.

[0050] Please see Figure 4 , Figure 4 A specific implementation of step S4 is shown below: S41: Apply sliding window linear regression to the variance, skewness, and kurtosis in the degradation feature sequence to obtain the variance slope, skewness slope, and kurtosis slope, respectively. S42: Calculate the comprehensive degradation score based on the variance slope, skewness slope, and kurtosis slope.

[0051] Specifically, sliding window linear regression is applied to the variance, skewness, and kurtosis in the degradation feature sequence to extract the trends of these statistical features over time. Sliding window linear regression is an effective time series analysis technique. Its core lies in moving a fixed-size window across the data sequence and performing linear regression analysis within each window to obtain a series of local trend information. For example, a window size can be set to contain statistical feature data points from the most recent N detection times. Then, least-squares linear regression is performed on these N data points to calculate the slope within the window. The window can slide forward with a step size of one data point or with a larger step size to balance computational efficiency and the granularity of trend capture. In this way, short-term fluctuations and noise can be effectively smoothed, thus revealing the long-term trend of battery degradation more accurately.

[0052] Through the sliding window linear regression process described above, we can obtain the variance slope, skewness slope, and kurtosis slope. These slopes are direct outputs of the linear regression model, quantifying the rate and direction of change of these three statistical characteristics over time. For example, a positive variance slope indicates that the dispersion of cell voltages within the battery pack is gradually increasing, leading to poorer consistency; a negative skewness slope may indicate a trend of voltage distribution shifting from one direction to another. These slope values ​​are dynamic and trend-based indicators, and compared to static statistical characteristics themselves, they are more sensitive to subtle changes in the consistency within the battery.

[0053] Based on this, the comprehensive degradation score is calculated using the variance slope, skewness slope, and kurtosis slope. The calculation of the comprehensive degradation score integrates the slope information from multiple dimensions into a single, comprehensive indicator that reflects the degree of battery degradation. For example, a weighted summation method can be used, assigning a preset weight coefficient to each slope (variance slope, skewness slope, kurtosis slope), and then linearly summing the weighted slopes to obtain the comprehensive degradation score. The weight coefficients can be determined based on the mechanism of battery degradation, expert experience, or optimized through machine learning methods to reflect the relative importance of changes in different statistical characteristics to battery degradation. Alternatively, a non-linear combination method can be used; for example, each slope can be input into a machine learning model such as a multilayer perceptron (MLP) or support vector machine (SVM), allowing the model to learn and output the comprehensive degradation score, thereby capturing the more complex non-linear relationships between the slopes.

[0054] The formula for calculating the overall degradation score is as follows: ; in, , , These are the weighting coefficients for each dimension. , , This is the normalized mapping function from trend values ​​to scores.

[0055] This application overcomes the limitations of relying solely on single statistical features or raw features by fusing these multi-dimensional slope changes into a comprehensive degradation score, achieving a comprehensive and in-depth correlation analysis of early battery degradation trends. This enables the early warning system to sensitively detect subtle, gradual changes in the consistency within the battery pack during the early stages of battery degradation—before traditional extreme indicators such as differential pressure reach warning thresholds. This significantly improves the accuracy and timeliness of early battery degradation warnings, providing a valuable window for battery maintenance and intervention.

[0056] Please see Figure 5 , Figure 5A specific implementation of step S42 is shown below: S421: The absolute values ​​of the variance slope, the skewness slope, and the kurtosis slope are normalized using an exponential saturation mapping function to obtain a normalized score.

[0057] Furthermore, the form of the exponentially saturated mapping function is as follows: ; in, The slope of the input change or its absolute value; The saturation rate coefficient has a value ranging from 5 to 20, with a preferred value of 10. This represents the inflection point, determined based on historical data, and is typically set to 0. The mapping function has the following characteristic: when... When f(x) approaches 0, When f(x) approaches 1, it achieves a sensitive response to small trends and a saturation limit on extreme values, avoiding single-point anomalies that cause scoring distortion.

[0058] For skewness slope Preserve symbol information to distinguish the direction of skewing: The output range is [0,1], when This indicates a worsening rightward bias (higher voltage in some cells). The leftward deviation is aggravated (the voltage of some cells is too low), and both directions are considered as potential degradation signals and are included in the scoring.

[0059] S422: The normalized scores are weighted and summed with preset weight coefficients to obtain the comprehensive degradation score.

[0060] Specifically, the exponential saturation mapping function used in this application is a nonlinear function characterized by its ability to map input values ​​from a relatively wide range to a finite output range, typically 0 to 1 or -1 to 1. The function curve is usually S-shaped; when the input value is in a specific sensitive region, the output value changes significantly, while when the input value is too large or too small, the output value tends to saturate and the change slows down. This characteristic makes it very suitable for normalizing the slopes of statistical features with different dimensions and ranges of variation, effectively suppressing the excessive influence of extreme values ​​on the overall score while maintaining the discriminative power in key variation regions. For example, the function can be designed so that after the input value reaches a certain threshold, the growth rate of the output value slows down and tends to stabilize, thereby avoiding excessive disturbance to the overall evaluation result by individual extreme data points. The purpose of normalization is to transform the values ​​of different statistical features (variance slope, absolute value of skewness slope, and kurtosis slope) to a uniform scale, eliminating the influence caused by differences in dimensions, numerical ranges, and distributions. Normalization ensures the comparability of various features in subsequent comprehensive score calculations, preventing features with large values ​​from dominating the scoring results. This allows for a fair measurement of the contribution of each feature to the degree of battery degradation. Normalized scores are quantified values ​​of the slopes of various statistical features within a uniform numerical range, obtained after processing with an exponential saturation mapping function. These scores reflect the relative intensity and direction of change of each statistical feature during battery degradation, and because they are on a uniform scale, they can be directly compared and weighted. Preset weight coefficients are numerical values ​​assigned to each normalized score to characterize the relative importance or sensitivity of that statistical feature in indicating early battery degradation. These weight coefficients can be set and adjusted based on battery type, operating environment, historical degradation data analysis, or expert experience. For example, machine learning analysis of large amounts of battery degradation data can determine the predictive power of different statistical features for early battery degradation, and then assign corresponding weights; or, based on the experience of battery engineers, higher weights can be given to features considered to reflect degradation earlier and more accurately. Weighted summation is a mathematical operation that linearly combines multiple normalized scores according to their preset weight coefficients. By multiplying each normalized score by its corresponding weighting coefficient and then summing all the products, a comprehensive numerical value can be obtained. This method can flexibly integrate degradation information from different dimensions, enabling the final comprehensive score to more comprehensively and accurately reflect the overall degradation status of the battery pack, while highlighting the impact of key degradation indicators. The comprehensive degradation score is a single value obtained after normalization and weighted summation. It comprehensively quantifies the combined trend of the dispersion, symmetry, and concentration of voltage distribution within the battery pack over time. As a core indicator of early battery degradation, this score can more sensitively capture subtle changes in battery pack consistency, providing a reliable quantitative basis for subsequent early warning level judgments.

[0061] S5: Compare the comprehensive degradation score with the preset grading threshold, and compare each of the change slopes with the single-dimensional trend threshold to determine and output the warning level.

[0062] Specifically, the system compares the overall degradation score with a preset grading threshold, and compares each slope of change with a single-dimensional trend threshold to determine and output the warning level. The calculated overall degradation score is compared with the preset grading threshold. For example, a single overall threshold can be set; when the overall degradation score exceeds this threshold, a warning is triggered. Simultaneously, the slope of change for each statistical feature is compared with its respective preset single-dimensional trend threshold. For example, a threshold can be set for the average slope, range slope, etc.; when any slope exceeds its corresponding threshold, a warning is also triggered. By comprehensively considering these two comparison results, the system can determine and output the corresponding warning level.

[0063] Please see Figure 6 , Figure 6 A specific implementation of step S5 is shown below: S51: If the overall degradation score is less than the first grading threshold, and each of the change slopes is lower than the single-dimensional trend threshold, then the warning level is determined as the normal level, and the normal level is output. S52: If the overall degradation score is greater than or equal to the first grading threshold and less than the second grading threshold, or any of the change slopes exceeds the single-dimensional trend threshold, then the warning level is determined as the attention level, and the attention level is output. S53: If the overall degradation score is greater than or equal to the second grading threshold and less than the third grading threshold, or at least two of the change slopes exceed the single-dimensional trend threshold, then the warning level is determined as the alert level, and the alert level is output. S54: If the overall degradation score is greater than or equal to the third grading threshold, then the warning level is determined as the severe alert level, and the severe alert level is output.

[0064] The preset grading thresholds are numerical boundaries used to quantify and distinguish the degree of battery degradation. These thresholds can be determined based on statistical analysis of a large amount of historical battery operating data, for example, by cluster analysis of the comprehensive degradation scores of batteries in different health states. Furthermore, these thresholds can also be set according to battery type, application scenario, or manufacturer's empirical standards to ensure they accurately reflect the actual degradation risk. Comparing the comprehensive degradation score with the preset grading thresholds aims to map the overall degradation risk of the battery (represented by the comprehensive degradation score) to a predefined risk range. Specifically, this can be achieved through a series of conditional statements (such as if-else structures), comparing the calculated comprehensive degradation score sequentially with the first, second, and third grading thresholds to initially determine its risk range. For example, if the comprehensive degradation score falls between the first and second grading thresholds, it indicates that the battery is in a medium-risk range. Comparing each of the aforementioned slopes with single-dimensional trend thresholds provides a more detailed, multi-dimensional assessment of the battery degradation trend. The slope of change for each statistical characteristic (such as variance, skewness, and kurtosis) corresponds to one or a set of single-dimensional trend thresholds. These thresholds can be determined based on the typical range of change of each statistical characteristic quantity during normal operation and early degradation stages. For example, by analyzing long-term monitoring data of healthy batteries, an upper limit for the allowable slope of change can be set. In practical applications, different single-dimensional trend thresholds can be set for variance slope, skewness slope, and kurtosis slope to capture abnormal changes in different dimensions. The step of determining and outputting the warning level is to determine and release the battery warning level based on the comparison results of the comprehensive degradation score and the grading threshold, as well as the comparison results of each slope of change and the single-dimensional trend threshold, through preset logical rules. The output form of the warning level can be a numerical code, text description (such as "normal", "attention", "warning", "serious warning"), or graphical indication, and presented through the vehicle display screen, remote monitoring platform, or battery management system interface. In a specific embodiment, the single-dimensional trend thresholds are set as follows: variance slope threshold is 0.000015, skewness slope threshold is 0.05, kurtosis slope threshold is 0.15, first grading threshold is 0.3, second grading threshold is 0.5, and third grading threshold is 0.75.

[0065] Specifically, if the overall degradation score is less than the first grading threshold, and each of the stated slopes is lower than the single-dimensional trend threshold, then the warning level is determined to be a normal level, and the normal level is output. This rule defines the criteria for determining whether a battery is in a healthy state. It requires that the overall degradation risk (overall degradation score) and the degradation trends (states of change) of all key statistical features are within a safe range. For example, when the overall degradation score is below 0.2, and the variance slope, skewness slope, and kurtosis slope do not exceed their respective single-dimensional trend thresholds, the system determines the battery to be at a normal level. If the overall degradation score is greater than or equal to the first grading threshold and less than the second grading threshold, or any stated slope exceeds the single-dimensional trend threshold, then the warning level is determined to be a concern level, and the concern level is output. This rule is used to identify situations where a battery begins to show slight signs of degradation. Even if the overall degradation score has not yet reached a high level, as long as the slope of change of one or more key statistical features begins to exceed its single-dimensional trend threshold, it indicates that the battery consistency may be undergoing subtle changes, requiring attention. For example, when the overall degradation score is between 0.2 and 0.5, or the variance slope exceeds its single-dimensional trend threshold, but the skewness slope and kurtosis slope are still within the threshold, the system can determine it as a concern level. If the overall degradation score is greater than or equal to the second grading threshold and less than the third grading threshold, or at least two of the change slopes exceed the single-dimensional trend threshold, then the warning level is determined as a warning level, and the warning level is output. This rule aims to identify situations where battery degradation further intensifies. At this point, the battery's overall degradation score has reached a medium-risk level, and the change slopes of at least two key statistical features have exceeded their single-dimensional trend thresholds, indicating a significant trend of deteriorating battery consistency, requiring preventative maintenance measures. For example, when the overall degradation score is between 0.5 and 0.75, or both the variance slope and skewness slope exceed their single-dimensional trend thresholds, the system can determine it as a warning level. If the overall degradation score is greater than or equal to the third grading threshold, then the warning level is determined as a severe warning level, and the severe warning level is output. This rule is used to identify situations where a battery is in a severely degraded state, potentially posing a high safety risk. In this case, the overall degradation score has reached the highest risk level, indicating that the battery's overall performance and consistency have been severely compromised, requiring immediate inspection or replacement. For example, when the overall degradation score exceeds 0.75, regardless of the specific details of each slope, the system directly classifies it as a severe warning level.

[0066] This application effectively addresses the problems of ambiguous warning level classification and difficulty in providing differentiated operation and maintenance guidance in existing technologies. By introducing a multi-level preset grading threshold system, it combines the macroscopic assessment of comprehensive degradation scores with the microscopic trend analysis of the slope of changes in various statistical characteristic quantities, achieving refined grading and early warning of battery degradation status. This grading logic not only improves the accuracy of early warnings but also provides more targeted reference for subsequent maintenance decisions through multi-dimensional status determination. For example, when the battery is at the "concern level," users can be advised to conduct preliminary checks or adjust charging habits; when it is at the "warning level," professional testing or replacement of some battery cells may be necessary; and the "serious warning level" directly triggers the highest level of safety response. This refined grading and early warning mechanism enables sensitive identification of abnormal trends in the early stages of battery degradation, before the differential pressure reaches traditional alarm thresholds. This effectively compensates for the lag and crudeness of traditional methods in early degradation identification, thereby enabling refined management of the entire battery lifecycle, extending battery life, and improving system safety.

[0067] This application provides a specific embodiment where, during the k-th periodic inspection of a vehicle equipped with a ternary lithium battery pack, the BMS reports the voltage values ​​of 100 individual battery cells. First, a voltage sample set is collected. 100 cell voltage values ​​are obtained, for example: V = {3.449, 3.458, 3.452, 3.463, 3.446, ...} (a total of 100 values). Then, a statistical characteristic is calculated: the sample mean. = 3.457V, variance σ² = 0.000065, skewness = 0.1619, kurtosis = 3.1778, maximum pressure drop = 45mV (still within the 50mV threshold). The collected data and historical data were combined to form a degradation feature sequence. The statistical characteristics of the vehicle from the previous four inspections have been recorded: Test 1: σ²=0.000020, skew=0.0500, kurt=2.9500, diff=8mV; Test 2: σ²=0.000035, skew=0.1200, kurt=3.1000, diff=12mV; Test 3: σ²=0.000058, skew=0.2200, kurt=3.4500, diff=15mV; Test 4: σ²=0.000065, skew=0.1619, kurt=3.1778, diff=45mV (this time); Next, the trend value is calculated. Linear regression is performed using the last four detection windows (window size w=4). The window size w is set according to the detection frequency, with 3 to 10 detections recommended, corresponding to a time span of 2 hours to 30 days. k_σ²=0.00002 (variance shows an increasing trend), k_skew= 0.04356 (skewness increases, indicating that the voltage of some cells deviates positively), k_kurt= 0.1034 (kurtosis changes, indicating a change in distribution pattern).

[0068] Input the above trend values ​​into the exponential saturation mapping function (take k=10, variance and skewness are set to...). kurtosis To accommodate its range of variation, a normalized score is obtained: =0.5000、 =0.6072、 =0.0490.

[0069] Calculate the overall degradation score. The weights are preset to α=0.4, β=0.3, γ=0.3. After normalization and mapping, we get: S=0.4×0.5000+0.3×0.6072+0.3×0.0490=0.3969.

[0070] Finally, the warning level is determined. In this embodiment, the first grading threshold S1 is 0.3, the second grading threshold S2 is 0.5, and the third grading threshold S3 is 0.75, which can be adjusted according to the actual battery type and operating conditions. S=0.3969 falls into [ =0.3, =0.5) range, output the level of concern. At this time, the maximum differential pressure is only 45mV, which would not trigger any alarm according to existing technology. However, the statistical characteristic analysis of this application has detected the degradation trend and issued an early warning.

[0071] Please refer to Figure 7 As a response to the above Figure 1 The implementation of the method shown in this application provides an embodiment of a battery early degradation early warning device, which is similar to... Figure 1 Corresponding to the method embodiments shown, the device can be specifically applied to various vehicles.

[0072] like Figure 7 As shown, the battery early degradation early warning device of this embodiment includes: a voltage sample set acquisition module 61, a statistical feature quantity calculation module 62, a degradation feature sequence construction module 63, a comprehensive degradation score calculation module 64, and an early warning level output module 65, wherein: The voltage sample set acquisition module 61 is used to acquire the voltage values ​​of all individual cells in the battery pack at the same time to form a voltage sample set. The statistical feature calculation module 62 is used to calculate multiple statistical features in the voltage sample set; The degradation feature sequence construction module 63 is used to construct a degradation feature sequence in chronological order for multiple statistical feature quantities obtained at different detection times for the same battery pack. The comprehensive degradation score calculation module 64 is used to extract the slope of change for each statistical feature quantity in the degradation feature sequence using sliding window linear regression, and to calculate the comprehensive degradation score based on the slope of change. The warning level output module 65 is used to compare the comprehensive degradation score with the preset grading threshold, and to compare each of the change slopes with the single-dimensional trend threshold, so as to determine and output the warning level.

[0073] Furthermore, the statistical characteristic calculation module 62 includes: The sample mean calculation unit is used to sum the voltage values ​​of all cells in the voltage sample set and divide by the total number of cells to calculate the sample mean. The feature calculation unit is used to calculate, based on the sample mean, the variance reflecting the degree of voltage dispersion, the skewness reflecting the symmetry of the voltage distribution, and the kurtosis reflecting the degree of concentration of the voltage distribution.

[0074] Furthermore, the comprehensive degradation score calculation module 64 includes: A linear regression unit is used to apply sliding window linear regression to the variance, skewness, and kurtosis in the degenerate feature sequence to obtain the variance slope, skewness slope, and kurtosis slope, respectively. The scoring calculation unit is used to calculate the comprehensive degradation score based on the variance slope, the skewness slope, and the kurtosis slope.

[0075] Furthermore, the scoring calculation unit includes: The normalized score generation subunit is used to normalize the variance slope, the absolute value of the skewness slope, and the kurtosis slope using an exponential saturation mapping function to obtain a normalized score. The weighted summation subunit is used to sum the normalized scores with preset weight coefficients to obtain the comprehensive degradation score.

[0076] Furthermore, the form of the exponentially saturated mapping function is as follows: ; in, The input is the slope of change or its absolute value. The saturation rate coefficient, This is the inflection point.

[0077] Furthermore, the preset grading thresholds include a first grading threshold, a second grading threshold, and a third grading threshold; the warning level output module 65 includes: The first level unit is used to determine the warning level as the normal level and output the normal level if the comprehensive degradation score is less than the first grading threshold and each of the change slopes is lower than the single-dimensional trend threshold. The second-level unit is used to determine the warning level as a concern level and output the concern level if the comprehensive degradation score is greater than or equal to the first grading threshold and less than the second grading threshold, or if any of the change slopes exceeds the single-dimensional trend threshold. The third-level unit is used to determine the warning level as an alert level and output the alert level if the comprehensive degradation score is greater than or equal to the second grading threshold and less than the third grading threshold, or if at least two of the change slopes exceed the single-dimensional trend threshold. The fourth level unit is used to determine the warning level as a severe warning level and output the severe warning level if the comprehensive degradation score is greater than or equal to the third level threshold.

[0078] Furthermore, the degradation feature sequence construction module 63 includes: A detection time stamping unit is used to mark the detection time for each group of the statistical feature quantities; A sequence construction unit is used to arrange historical detection records of the same battery pack in ascending order according to the detection time to construct the degradation feature sequence.

[0079] To address the aforementioned technical problems, embodiments of this application also provide a vehicle. Please refer to the following for details. Figure 8 , Figure 8 This is a basic structural block diagram of the vehicle in this embodiment.

[0080] Vehicle 7 includes a memory 71, a processor 72, and a network interface 73 that are interconnected via a system bus. It should be noted that... Figure 8 The vehicle 7 shown only has three components: memory 71, processor 72, and network interface 73. However, it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0081] The memory 71 includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 71 may be an internal storage unit of the vehicle 7, such as the hard disk or memory of the vehicle 7. In other embodiments, the memory 71 may also be an external storage device of the vehicle 7, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the vehicle 7. Of course, the memory 71 may also include both internal storage units and external storage devices of the vehicle 7. In this embodiment, the memory 71 is typically used to store the operating system and various application software installed on the vehicle 7, such as the program code of the early battery degradation warning method. In addition, the memory 71 may also be used to temporarily store various types of data that have been output or will be output.

[0082] In some embodiments, processor 72 may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip. This processor 72 is typically used to control the overall operation of vehicle 7. In this embodiment, processor 72 is used to run program code stored in memory 71 or process data, for example, to run the program code of the aforementioned early battery degradation warning method to implement various embodiments of the early battery degradation warning method.

[0083] The network interface 73 may include a wireless network interface or a wired network interface, which is typically used to establish a communication connection between the vehicle 7 and other electronic devices.

[0084] This application also provides another embodiment, namely, a computer-readable storage medium storing a computer program that can be executed by at least one processor to cause the at least one processor to perform the steps of the battery early degradation warning method described above.

[0085] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods of the various embodiments of this application.

[0086] Obviously, the embodiments described above are merely some embodiments of this application, not all embodiments. The accompanying drawings show preferred embodiments of this application, but do not limit the scope of this application. This application can be implemented in many different forms; rather, these embodiments are provided to provide a more thorough and comprehensive understanding of the disclosure of this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this application's specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the scope of protection of this application.

Claims

1. A method for early warning of battery degradation, characterized in that, include: Obtain the voltage values ​​of all individual cells in the battery pack at the same time to form a voltage sample set; Calculate multiple statistical features in the voltage sample set; For multiple statistical features obtained at different detection times for the same battery pack, a degradation feature sequence is constructed in chronological order. The slope of change of each statistical feature in the degradation feature sequence is extracted by sliding window linear regression, and a comprehensive degradation score is calculated based on the slope of change. The comprehensive degradation score is compared with a preset grading threshold, and each of the change slopes is compared with a single-dimensional trend threshold to determine and output the warning level. The calculation of multiple statistical features in the voltage sample set includes: The summation of all cell voltage values ​​in the voltage sample set is divided by the total number of cells to calculate the sample mean. Based on the sample mean, the variance reflecting the degree of voltage dispersion, the skewness reflecting the symmetry of the voltage distribution, and the kurtosis reflecting the degree of concentration of the voltage distribution are calculated. The step of extracting the slope of change for each statistical feature in the degradation feature sequence using sliding window linear regression, and calculating a comprehensive degradation score based on the slope of change, includes: Applying sliding window linear regression to the variance, skewness, and kurtosis in the degraded feature sequence yields the variance slope, skewness slope, and kurtosis slope. The comprehensive degradation score is calculated based on the variance slope, the skewness slope, and the kurtosis slope. The calculation of the comprehensive degradation score based on the variance slope, the skewness slope, and the kurtosis slope includes: The variance slope, the absolute value of the skewness slope, and the kurtosis slope are normalized using an exponential saturation mapping function to obtain a normalized score. The normalized scores are weighted and summed with preset weighting coefficients to obtain the comprehensive degradation score.

2. The method for early warning of battery degradation according to claim 1, characterized in that, The form of the exponential saturation mapping function is as follows: ； in, The input is the slope of change or its absolute value. The saturation rate coefficient, This is the inflection point.

3. The method for early warning of battery degradation according to any one of claims 1 to 2, characterized in that, The preset grading thresholds include a first grading threshold, a second grading threshold, and a third grading threshold; the step of comparing the comprehensive degradation score with the preset grading thresholds, and comparing each of the change slopes with a single-dimensional trend threshold, to determine and output the warning level, includes: If the comprehensive degradation score is less than the first grading threshold, and each of the change slopes is lower than the single-dimensional trend threshold, then the warning level is determined as the normal level, and the normal level is output. If the comprehensive degradation score is greater than or equal to the first grading threshold and less than the second grading threshold, or if the slope of a single change exceeds the single-dimensional trend threshold, then the warning level is determined as the attention level, and the attention level is output. If the comprehensive degradation score is greater than or equal to the second grading threshold and less than the third grading threshold, or if at least two of the change slopes exceed the single-dimensional trend threshold, then the warning level is determined as an alert level and the alert level is output. If the comprehensive degradation score is greater than or equal to the third grading threshold, the warning level is determined as a severe warning level, and the severe warning level is output.

4. The method for early warning of battery degradation according to any one of claims 1 to 2, characterized in that, The process of constructing a degradation feature sequence in chronological order for multiple statistical features obtained at different detection times for the same battery pack includes: The detection time is marked for each set of statistical features; The historical detection records of the same battery pack are arranged in ascending order according to the detection time to construct the degradation feature sequence.

5. A battery early degradation early warning device, characterized in that, include: The voltage sample set acquisition module is used to acquire the voltage values ​​of all individual cells in the battery pack at the same time to form a voltage sample set. The statistical feature calculation module is used to calculate multiple statistical features in the voltage sample set; The degradation feature sequence construction module is used to construct a degradation feature sequence in chronological order for multiple statistical feature quantities obtained at different detection times for the same battery pack. The comprehensive degradation score calculation module is used to extract the slope of change for each statistical feature quantity in the degradation feature sequence using sliding window linear regression, and to calculate the comprehensive degradation score based on the slope of change. The warning level output module is used to compare the comprehensive degradation score with the preset grading threshold, and to compare each of the change slopes with the single-dimensional trend threshold, so as to determine and output the warning level. The statistical characteristic calculation module includes: The sample mean calculation unit is used to sum the voltage values ​​of all cells in the voltage sample set and divide by the total number of cells to calculate the sample mean. The feature quantity calculation unit is used to calculate the variance reflecting the degree of voltage dispersion, the skewness reflecting the symmetry of voltage distribution, and the kurtosis reflecting the degree of concentration of voltage distribution based on the sample mean. The comprehensive degradation score calculation module includes: A linear regression unit is used to apply sliding window linear regression to the variance, skewness, and kurtosis in the degenerate feature sequence to obtain the variance slope, skewness slope, and kurtosis slope, respectively. The scoring calculation unit is used to calculate the comprehensive degradation score based on the variance slope, the skewness slope, and the kurtosis slope; The scoring calculation unit includes: The normalized score generation subunit is used to normalize the variance slope, the absolute value of the skewness slope, and the kurtosis slope using an exponential saturation mapping function to obtain a normalized score. The weighted summation subunit is used to sum the normalized scores with preset weight coefficients to obtain the comprehensive degradation score.

6. A vehicle, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the battery early degradation warning method as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the battery early degradation warning method as described in any one of claims 1 to 4.

Images